基于生物信息学构建卵巢浆液性腺癌预后风险模型及潜在中药预测分析

Construction of Prognostic Risk Model for Ovarian Serous Adenocarcinoma based on Bioinformatics and Predictive Analysis of Potential Traditional Chinese Medicine

目的:构建卵巢浆液性腺癌(Ovarian serous cystadenocarcinoma,OV)预后风险模型,针对核心预后基因筛选潜在中药。方法:数据来源于TCGA数据库,采用差异分析与单因素Cox回归筛选OV差异预后基因;以Kaplan-Meier(KM)曲线P<0.05,且受试者工作特征曲线(Receiver Operating Characteristic curve,ROC),以AUC>0.7为筛选条件,进一步聚焦到OV核心差异预后基因;采用单因素、多因素Cox回归构建卵巢浆液性腺癌核心差异预后基因风险评分系统,并结合临床信息开发临床预测模型,对预测模型进行Calibration分析与DCA分析。通过 CIBERSORT 反卷积法分析 24 类免疫细胞在OV中的含量及比例。通过Coremine、TCMSP、TCMIP数据库预测OV差异核心预后基因的潜在中药。结果:TCGA数据库获得346个DEGs、14个OV差异预后基因、7个OV核心差异预后基因达。非配对t检验显示FGF20、TRPM3、KLK3在卵巢浆液性腺癌中低表达,SAA2-SAA4、ADGRG5、EPB41L1、DAP1在卵巢浆液性腺癌中高表达。7个核心差异预后基因多因素Cox回归分析发现,SAA2-SAA4、ADGRG5、TRPM3、DAPL1(P<0.05)均可作为OV的独立预后基因。7个核心差异预后基因构建的风险评分系统能显著区分生存,低风险评分具有的生存预后,是卵巢浆液性腺癌的危险因素,time-ROC提示风险评分能较好的评估生存预后,且5年的预测能力最佳。临床预测模型诺莫图可有效预测1年、3年、5年的生存概率。免疫浸润结果构建了与24种免疫细胞的相关性。coremine数据库预测出三七叶、三七花、地榆等18个OV差异核心预后基因的潜在中药,药物多归肝经,功效多与补虚止血有关。结论:本研究筛选出OV核心差异预后基因,构建了OV预后风险评分系统,并开发了临床预后模型诺莫图,同时对OV核心差异预后基因进行了免疫浸润分析,预测了治疗OV的潜在中药,为中医药干预卵巢癌提供了新思路。

Objective:To construct a prognostic risk model for ovarian serous cystadenocarcinoma (OV),and to screen potential traditional Chinese medicines based on core prognostic genes.Methods:The data were obtained from the TCGA database,and differential analysis and univariate Cox regression were used to screen OV differential prognostic genes;Kaplan Meier (KM) curve P0.7 as the screening condition,further focus on the core differential prognostic genes of OV;using univariate and multivariate Cox regression to construct the core differential prognostic gene risk scoring system for ovarian serous adenocarcinoma,and combining clinical information to develop a clinical prediction model.Perform Calibration analysis and DCA analysis.The content and proportion of 24 types of immune cells in OV were analyzed by CIBER SORT deconvolution method.Potential traditional Chinese medicines for predicting OV differential core prognostic genes through Coremine,TCMSP,TCMIP databases.Results:346 DEGs,14 OV differential prognostic genes,and 7 OV core differential prognostic genes were obtained from the TCGA database.Unpaired t-test showed that FGF20,TRPM3,and KLK3 were lowly expressed in ovarian serous adenocarcinoma,and SAA2-SAA4,ADGRG5,EPB41L1,and DAP1 were highly expressed in ovarian serous adenocarcinoma.Multivariate Cox regression analysis of seven core differential prognostic genes found that SAA2-SAA4,ADGRG5,TRPM3,DAPL1 (P<0.05) could be used as independent prognostic genes for OV.The risk scoring system constructed by 7 core differential prognostic genes can significantly distinguish survival.Low risk score has better survival prognosis and is a risk factor for ovarian serous adenocarcinoma.Time-ROC suggests that risk score can better evaluate survival prognosis,and 5 years the best predictive ability.The clinical prediction model Nomomap can effectively predict 1-year,3-year,and 5-year survival probability.The immune infiltration results were constructed to correlate with 24 types of immune cells.The coremine database predicted the potential Chinese medicines of 18 core prognostic genes with OV differences,such as Panax notoginseng leaves,Panax notoginseng flowers,and Diyu,etc.Most of the drugs belong to the liver meridian,and their efficacy is mostly related to tonic deficiency and hemostasis.Conclusion:This study screened out the OV core differential prognostic genes,constructed the OV prognostic risk scoring system,and developed a clinical prognostic model Nomo map.At the same time,the immune infiltration analysis of the OV core differential prognostic genes was performed to predict the potential Chinese medicines for the treatment of OV.,which provides a new idea for Chinese medicine to intervene in ovarian cancer.